氨基酸是含有碱性氨基和酸性羧基的有机化合物, 是构成蛋白质的基本单位, 其种类, 数量和排列直接影响蛋白质的生物功能, 对维持机体功能有重要意义。 氨基酸分子间振动模式(扭转, 氢键和集体振动)大部分处于太赫兹(THz)波段, 表现出独特的吸收特征, 因此, 对氨基酸进行THz光谱研究, 能够更全面了解生物特性。 总结前人实测的亮氨酸与异亮氨酸位于0.2~2.6 THz波段的吸收谱, 同时, 利用量子化学计算方法解释其形成机理。 使用Gaussian09软件对单分子构型模拟计算, 模拟方法为半经验法(PM6), 从头计算法(HF, MP2)和密度泛函理论(B3LYP, M06-2X)结合6-311+G(d, p)高斯型基组; 使用Materials Studio 2019软件对晶胞构型模拟计算, 模拟方法为广义梯度近似的PBE, PBEsol, RPBE和WC等四种密度泛函结合平面波基组。 结果表明: 单分子构型模拟均缺少吸收峰位, 不同方法对同一振动模式的峰位计算不同, 因此, 对分子间相互作用较强的结构, 进行单一方法的该构型模拟, 很大程度不能正确匹配振动模式, 且受原子轨道线性组合方法影响, 与输入结构相比, 输出结构由COO^-和$NH_{3}^{+}$基团变为COOH和NH₂, 无法体现实际振动模式; 晶胞构型模拟对分子内和分子间振动模式描述, 吸收峰位与实测值匹配较好, 不存在质子转移情况, 较好指认实测峰位的振动模式。 亮氨酸与异亮氨酸使用PBEsol泛函计算结果最接近实测值, 说明模拟计算需充分考虑结构与泛函的匹配性, 即对结构交换关联能的描述, 也说明同一泛函对异构体的普适性, 此外, 不能以结构优化后差异作为判断泛函是否适用的标准。 晶胞构型计算结果包含分子间振动模式, 是单分子构型无法得到的结果, 且数据进行半峰全宽拟合, 导致两种构型结果在某一实测峰位处的振动模式存在差异。

As organic compounds containing an alkaline amino group and acidic carboxyl group, an amino acid is the basic unit of protein, whose type, quantity and arrangement directly affect the biological function of protein with great significance to maintain the body function. Most amino acid intermolecular vibrational modes (torsion, hydrogen bonding and collective vibrations) occur in the THz band and exhibit unique absorption characteristics. Consequently, THz spectroscopic studies of amino acids can give a more comprehensive understanding of biological properties. It was a summary of the absorption spectra of leucine and isoleucine located in the 0.2~2.6 THz band measured by previous authors, and, at the same time, the formation mechanism was explained using quantum chemical calculations. Using Gaussian09 software for single-molecule configuration simulations, the simulations were performed by semi-empirical method (PM6), ab initio method (HF, MP2) and density functional theory (B3LYP, M06-2X) combined with 6-311+G(d,p) Gaussian basis groups. Materials Studio 2019 software was used to calculate the cell configuration simulations for four density generalized gradient approximations such as PBE, PBEsol, RPBE and WC combined with plane wave basis groups. The results indicated that the single-molecule configuration simulations lacked absorption peak positions, and the peak positions were calculated differently for the same vibrational mode by different methods. So, for structures with strong intermolecular interactions, the simulation of this configuration by a single method did not correctly match the vibrational modes to a large extent and was influenced by the linear combination of atomic orbitals method. Compared to the input structure, the output structure changed from COO^- and $NH_{3}^{+}$ groups to COOH and NH₂, which did not reflect the actual vibrational mode. The intra and intermolecular vibrational modes were described by the cell configuration simulation. The absorption peak positions matched well with the measured values without proton transfer. The vibrational modes of the measured peak positions were better identified. The closest results of leucine and isoleucine calculations using the PBEsol general function to the measured values indicated that full consideration needs to be given to the matching of structure and general function in the simulation, i.e., the description of the structure exchange association energy and showed the universality of the same general function for the isomers. In addition, the difference after structural optimization cannot be used as the criterion to judge functional applicability. The results of cell configuration calculation include the intermolecular vibration mode, which cannot be obtained with a single molecular configuration. Moreover, the full width at half maximum fitting leads to the difference in the vibration mode of the two configuration results at a measured peak position.